Hi,
I'm facing a problem with the Outlier Removal node(under community nodes group)... when I launch it with 1 mln of rows all went fine. Just 1 minute of execution and all went fine.
But if I launch it for example with more than 4/5 mln of rows everything went slow and Knime crashes after I run the node with this error:
Execute failed: GC overhead limit exceeded
Before executing I have also received this WARNINGS:
Could not load property 'SettingsModelBoolean ('inplace')'of node OutlierRemoval
WARN Outlier Removal 0:44 Could not load property 'SettingsModelBoolean ('inplace')'of node OutlierRemoval.
Please re-execute the node to get rid of this problem!
WARN Outlier Removal 0:44 Could not load property 'SettingsModelString ('group.wells.by')'of node OutlierRemoval
WARN Outlier Removal 0:44 Could not load property 'SettingsModelString ('group.wells.by')'of node OutlierRemoval.
Please re-execute the node to get rid of this problem!
The strange fact is that if I try to split the data in 4 partitions (each of them made of 1.5 mln rows) and I execute 4 Outlier Removal node then it works. And after that I concatenate all the parts together with the union of column. But this is very combersome and expensive as resources and execution time. And also I'm not quite sure is a good approach... and the most important limit is that I need to execute manually one node at time cause If I execute everything together it crashes again.
I'm currently on Ubuntu, 4 cores and 16 gb of ram.
This is my actual knime.ini configuration:
-startup plugins/org.eclipse.equinox.launcher_1.3.200.v20160318-1642.jar --launcher.library plugins/org.eclipse.equinox.launcher.gtk.linux.x86_64_1.1.400.v20160518-1444 -vmargs -server -Dsun.java2d.d3d=false -Dosgi.classloader.lock=classname -XX:+UnlockDiagnosticVMOptions -XX:+UnsyncloadClass -Dsun.net.client.defaultReadTimeout=0 -XX:CompileCommand=exclude,javax/swing/text/GlyphView,getBreakSpot -Xmx10g -Dorg.eclipse.swt.internal.gtk.disablePrinting
Someone can give me an help?
Thanks in advice.